Jeroen Groot

Agricultural production in Northern Ghana is dominated by smallholder farm systems, which are characterized by low inputs and low outputs, declining soil fertility, large yield gaps and limited adoption of agricultural technologies. There is an urgent need for alternative farm designs that are more productive, yet more sustainable. Technology packages for sustainable intensification are promoted by an R4D project in the Upper East, Upper West and Northern Regions of Ghana. In this paper, we analyse differences in perceived suitability, and modelled technical impact per technology package. We used a locally validated framework to categorise farm systems diversity that considers both, the horizontal (between households) and vertical (within households) dimension of diversity. Farm households were classified along a gradient of resource endowment. We selected one representative farm per type and per region to assess and compare their socioeconomic and environmental performance (farm profitability, labour and soil organic matter inputs) using the whole-farm model Farm DESIGN. We then used Farm DESIGN to assess the potential impact of five proposed technology packages and to explore promising alternative farm configurations. We discussed model assumptions and results with farmers, including alternative cropping patterns and trade-offs. We evaluated the packages with different household members using a weighted scoring technique, subsequently juxtaposing model results with farmer perceptions. Large differences prevailed among and within farms per type and per region, with low resource endowed farms being projected to benefit most in relative and least in absolute terms from an adoption of the packages. Farmer feedback confirmed the accuracy of alternative farm configurations, as determined by the model. However, the feedback also revealed that the most profitable farm designs would be hard to attain in reality, particularly for members of low and medium resource endowed households, due to high initial investment costs. Within households, women were more positive about the packages than men, since men heavily penalized extra costs and labour, translating into a greater congruence of model results with the male evaluation. We discuss the importance of distinguishing between technical (technology i.e. purchased tools and inputs) and managerial (techniques e.g. row planting) package components. We conclude that operationalizing inter-and intra-household diversity is a fundamental step in identifying sensible solutions for the challenges smallholder farm systems face in Northern Ghana.

1. Introduction
The first Green Revolution, dating from the 1960s, contributed to large increases in agricultural production in Asia and Latin America, but largely bypassed Africa. This model generated systems geared towards high productivity, and maintained in that state through large and regular inputs of energy in the form of mechanical operations, irrigation, and application of fertilizers and other agrochemicals. These systems are only viable in circumstances where energy is cheap and/or subsidized. As the world enters an era of energy scarcity, the pressing calls for a second Green Revolution in Africa will have to be answered by a different model. Natural ecosystems are maintained in a productive state through self-organization, i.e. complex internal flows of energy between a large diversity of components. The objective of this study was to observe the state of energy (efficiency) and explore the ecosystem (flow) properties of farms as a potential for sustainable intensification in Southern Ethiopia.

2. Materials and methods
The study was carried out in Hawassa lake region, Southern Ethiopia. Based on a survey made by Yodit Kebede (2013) to 173 farmers made in three sites differing on their perennial-annual crop composition, twelve farms were selected as study cases. Data was collected by farm visits, interviews (resource flow mapping, labour calendars) and complemented by empirical measurements of biomass flows and standing biomass and focus group discussions. Coefficients of energy use for labour were recalculated based on the stress scores and maximum and minimum values found on literature (Vigne et al., 2012). All materials and labour were given an energy value in MegaJoules (MJ) using energy coefficients from different sources (ILRI database, 2011, Vigne et al., 2012, Feedipedia, 2012).The first energy analysis visualizes each farm as a “black-box” and takes into account inputs –I-  (labour, fertilizers, external biomass) and outputs –O- (agricultural production) to calculate efficiencies (O/I) or total balance (O-I) (Funes-Mozonte et al., 2008). Labour from crops, animal caring and household tasks were considered in the “black-box” analysis.
Secondly, the household-farming systems were conceptualized as trophic webs: elements (household and farm elements) interconnected by the internal and external flows of energy, including biomass, labour and energy cost of inputs. Ecological Network Analysis (ENA) was used to assess the interactions between system elements and provide an overview the internal functioning and ecological (flow) properties of the agroecosystems (Alvarez et al., 2009, Ulanowicz, 2004, Rufino et al., 2009).

3. Result and discussion
3.1. Energy use efficiency
It was found that energy use efficiency (EUE) tended to decline with increasing inputs (I) (Fig 1b), although outputs (O) increased with greater I (Fig 1a). Higher energy input can increase productivity, however this relationship is not proportional, which means that as inputs increase the energy use efficiency will decrease (Gliessman, 1998). The average EUE for all farms was 1.7. Funes-Mozonte et al. (2008) designed integrated crop-livestock systems in Cuba, achieving EUE values of 9.6 and 9.8, meaning there is room for improvement (or intensification) in the studied systems.

3.2. System stability through internal energy cycling
Analyzing the energy flows of a system through ENA allows observing different system properties that in other case would not be perceived. Three important indicators are presented next. The internal capacity of the system (Ci) measures the diversity of the interaction between elements (internal flows of energy). Redundance (R) can be understood as the “power on reserve” of a system to tolerate stress or changes. R can be explained as the availability of possible paths of energy in case one element or link would disappear, which we measure as the percentage of the flow diversity that is redundant – “realized redundancy”). Dependency (D) is the proportion of the energy flowing through the system that is imported from the environment (Ullanowicz, 2000, Alvarez et al., 2013).
It was found that as the internal capacity increases, the dependency on external inputs decreases (Fig 2a). On the other hand, redundance increases with an increasing internal capacity (Fig 2b). Therefore, a high internal capacity would at the same time reduce the need for energy inputs (D) and promote self-organization properties (R).
In Fig. 3 represents how the high input conventional systems are maintained in a productive state through input of energy in the form of mechanical operations, irrigation, and application of fertilizers and other agrochemicals. In opposition, natural ecosystems are maintained in a productive state through self-organization, i.e. complex internal flows of energy between a large diversity of components. ‘Agroecological systems’ mimic natural ecosystems by relying on self-organization and therefore requiring a low energy input. 

3.3. Promoting internal flows of energy for sustainability
The increase of internal capacity could be promoted as a sustainable intensification in order to minimize dependency and promote self-organization. But how to do it?
It was observed that internal capacity increases with a higher total number of crop and livestock species (Fig 4a) and with livestock density (kg ha-1) (Fig 4b). A higher crop and livestock diversity increases both the number of system elements and of internal interactions and provides room for both a higher internal capacity and redundance. A higher livestock density increases the internal capacity by promoting more energy exchanges within the system elements. Livestock allow to utilize resources that in other case would not be used (e.g. crop residues, trees) to produce new products to be used on farm (milk, meat, manure, labour) or exported.

4. Conclusion
A greater flow diversity should be targeted for sustainable intensification of agroecosystems in order to decrease dependency on external inputs and promote the capacity to withstand perturbations (redundance). This can be promoted by increasing the total number of crops and livestock and by including livestock/increasing its density. There is a great potential of using ecology knowledge about ecosystems in the agricultural context for sustainable intensification.

In this paper a set of criteria is proposed for the evaluation of the potential contribution of modelling tools to strengthening the multifunctionality of agriculture. The four main areas of evaluation are (1) policy relevance, (2) the temporal resolution and scope, (3) the degree to which spatial and socio-institutional scales and heterogeneity are addressed and (4) the level of integration in the assessment of scientific dimensions and of the multiple functions of agriculture. The evaluative criteria are applied to the portfolio of modelling approaches developed and applied in a joint project of the French research institute INRA and the Dutch Wageningen University & Research Centre. The CLUE-S model focuses on prediction of changes in multifunctional land-use at regional scale, given a set of predetermined scenarios or policy variants, e.g. for ex-ante policy assessment and initiation of discussions on regional development. The two other modelling approaches are complementary and aim to address multifunctional farming activities. The Landscape IMAGES framework generates a range of static images of possible but sometimes distant futures for multifunctional farming activities in a small region or landscape. It supports the exploration of trade-offs between financial returns from agriculture, landscape quality, nature conservation and restoration, and environmental quality. Co-Viability Analysis generates trajectories of states and farming decisions fulfilling a given set of ecological and productive constraints representing a desired and sustainable future. The three modelling approaches differ in their policy relevance, in the ways that spatial and socio-institutional scales are addressed and in their degree of explicitation of interaction between the various functions of agriculture, but jointly cover most of the desired capabilities for assessment of multifunctionality. Caveats were particularly identified in the integration of the socio-institutional dimension and the related heterogeneity. Although the model portfolio did not completely satisfy the demands of the set of evaluative criteria, it is concluded that, due to their complementarities, in combination the three models could significantly contribute to further development and strengthening of multifunctionality.